grpo-rl-training

Community

GRPO/RL training patterns

Authorovachiever
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Expert patterns and templates for Group Relative Policy Optimization (GRPO) training using TRL, enabling robust reward design and production-ready workflows.

Core Features & Use Cases

  • Production-ready templates for GRPO training (LoRA, reward composition, evaluation)
  • Reward function library with multiple reward shapes (format, correctness, style)
  • End-to-end workflow guidance from dataset prep to deployment

Quick Start

Use the production-ready template at templates/basic_grpo_training.py and customize rewards for your task.

Dependency Matrix

Required Modules

transformers>=4.47.0trl>=0.14.0datasets>=3.2.0peft>=0.14.0torch

Components

references

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: grpo-rl-training
Download link: https://github.com/ovachiever/droid-tings/archive/main.zip#grpo-rl-training

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.